Zebrafish (Danio rerio) is a tropical cyprinid that belongs to the teleost family. The heart of teleost has considerable species variability with respect to heart mass, gross morphology, histology, and vascularity (Santer, 1985). The ventricle varies from a single-layer trabeculation with no compact layers or coronary vessels in the myocardium, to a complicated myocardium with extensive trabeculation and coronary circulation (Farrell and Jones, 1992). The zebrafish has become an attractive model for molecular biologists and geneticists, with most studies focused on the early embryonic stages (Chen et al., 1996; Weinstein and Fishman, 1996; Alexander and Stainier, 1999). Our previous study of the zebrafish focused on the early heart development (Hu et al., 2000). Thus, information related to the morphology and function of the mature zebrafish heart, including the morphology of the compact layer and trabeculation, cellular organization, presence of the coronary vessels, the definition of the tension apparatus in the atrioventricular valve, or functional aspects (relationship of the atrial, ventricular, dorsal, and ventral aortic pressures) of the zebrafish heart, is still needed.
The formation of the heart structure and the properties of cellular organelles are important in the assessment of genetic models. Thus, we provide a detailed description of the morphologic structure and hemodynamic function of the mature heart. These data serve as a framework for structure-function analysis of the zebrafish cardiovascular system.
MATERIALS AND METHODS
Scanning and Transmission Electron Microscopy
Adult zebrafish hearts (3 months postfertilization) were perfusion-fixed in diastole at high flow low pressure (Pexieder, 1981) with 2.5% glutaraldehyde-2% paraformaldehyde in isotonic 0.1 M cacodylate buffer, and 4 × 10−4 g/kg dose of ditilazem for 4 hr. After the cacodylate buffer wash, the hearts (n = 20) were examined for their external morphology, including venous and arterial system in the heart region, branchial vessels, ductus Cuvier, sinus venosus, atrium, ventricle, and bulbus arteriosus.
For scanning electron microscopy, the hearts (n = 40) were cut with microscissors in frontal, transverse, or sagittal sections. The specimens were post-fixed in 2% osmium tetraoxide, then in a saturated solution of thiocarbohydrazide to permeate and harden the tissue. The sections were dehydrated through ethanol series, and critical point dried with hexamethyldisilazane. After mounting on stubs with conductive carbon adhesive tabs, the tissues were coated with gold-palladium in a sputter coater (Technics, Springfield, VA), then examined and photographed in a Hitachi S450 scanning electron microscope (Mountain View, CA).
For transmission electron microscopy, the hearts (n = 40) were post-fixed in 2% osmium tetroxide, dehydrated, and embedded in Spurr's epoxy resin. Randomized thin sections were cut from each ventricle, atrium, and bulbus arteriosus, stained with uranyl acetate, then lead citrate, and photographed in a Hitachi H7100 transmission electron microscope (Mountain View, CA). Final magnification of the photomicrographs was 6,200×, 12,600×, and 18,200×.
Two sets of photomicrographs (6,200× and 18,200×) for each atrium and ventricle per heart were scanned digitally at a resolution of 1,200 dpi with a HP ScanJet ADF scanner (Hewlett-Packard, Palo Alto, CA) to the 733 MHz PC. The myocyte cytoplasmic area (6,200×) was traced by using SigmaScan Pro morphometric analysis software (SPSS, Chicago, IL). Fractional myocardial cytoplasmic volume of myofibrils and mitochondria (18,200×) was determined by point counting techniques (Weibel, 1979; Williams, 1981; Clark et al., 1986, 1989) with the identical software. Organelles other than those mentioned, including the ground substance but excluding nuclei and vacuoles, were categorized as cytoplasm. The overlay grid had at least 1,120 intercepts (points) after the photomicrograph resized to 2,300 × 1,800 pixel. A total of 40 micrographs (20 each from the atrium and ventricle) were accounted for the point counting. The number of intercepts falling on a given structure (Pi) and the total number of points (Pcytoplasm) were related to the relative volume fraction (Vi) of the structure per total cytoplasmic volume (Vcytoplasm) as shown: Vi/Vcytoplasm = Pi/Pcytoplasm.
Adult zebrafish hearts were perfusion-fixed in similar manner as scanning and transmission electron microscopy, except with 10% paraformaldehyde in isotonic phosphate-buffered solution. The hearts (n = 20) were dehydrated through an ascending ethanol series, and embedded in paraffin in transverse, frontal, and sagittal orientation. Serial sections were cut and stained with hematoxylin-eosin, or with Masson's trichrome stains to differentiate between collagen and muscle fibers (Carson, 1990). Images were captured digitally on a Zeiss Axioplan D7082 microscope fitted with an epifluorescence attachment and Axiophot ProgRes 3012 digital color camera (Carl Zeiss, Germany), and further processed by using Adobe Photoshop software (Adobe, San Jose, CA).
Zebrafish were anesthetized with tricaine methanesulfonate (MS222) at a dose of 0.40 g/kg in 0.075 g/L instant ocean solution (Hu et al., 2000) and transferred to a shallow concave trough of a custom-built glass chamber (Radnoti, Monrovia, CA). The trough was filled with instant ocean solution, and the temperature of the solution was monitored by a thermistor and maintained by 15 L/min flow of 30°C water through the hollow wall of the glass chamber. We measured the pressures in the atrium, ventricle, dorsal, and ventral aorta of the adult zebrafish with a model 900A servonull pressure system (World Precision Instruments, Sarasota, FL). The probe was a 5-μm diameter tip glass micropipette filled with 1 M NaCl. The cannula was zeroed at the level of the heart before and after pressure measurements.
The servonull measured pressure is linear (y = 0.995x − 0.23) and highly correlated (r = 0.99, S.E.E. = 0.11 mmHg) when compared with a standing water column over the range of 0 to 30 mmHg (Clark and Hu, 1982, 1990; Hu and Clark, 1989). Analog pressure waveforms were digitally sampled at 2-msec intervals by an analog to digital conversion board (LabView, National Instruments, Austin, TX) and stored for processing on Jaz cartridges (Iomega, Roy, UT). Ventricular dP/dt was mathematically derived with the analysis software. At least five consecutive cycles were analyzed for each measurement. Quantitative data (n ≥ 5) are presented as mean ± standard error of the mean, and analyzed by Student's t test, with statistically significant differences defined as a P-value of less than 5%.
The zebrafish heart in the pericardial sac was positioned anteroventrally to the thoracic cavity between the operculum and the pectoral bone of the pectoral fin. Blood returned to the heart from anterior and posterior cardinal veins, and from hepatic portal veins, which opened directly into sinus venosus (Fig. 1). The blood drained into the triangular single-chambered atrium, which encircled the dorsal side of the pyramidal ventricle. The single-chambered ventricle pumped the blood through the bulboventricular orifice into the pear-shaped bulbus arteriosus, which was interposed between the ventricle and ventral aorta. The ventral aorta was a cul-de-sac, cylindrical vessel, with four pairs of afferent branchial arteries branching dorsally on each side, and connected to the afferent branchial arterioles in the gill filaments under the operculum. Blood flowed from the afferent branchial arterioles of the gill filaments to the efferent branchial arteries. The first pair of efferent branchial arteries arched anterolaterally to the cephalic circulation. All four pairs of efferent branchial arteries converged medially to form the dorsal aorta.
The ventricle was bound externally by epicardium and internally by endocardium. In between the epicardial and endocardial cell layers were the muscular trabeculae that spanned the entire ventricle to form a spongy network of trabeculated layer (Fig. 2A). Apart from the complex mesh of trabeculation, there were fine trabeculae perpendicular to the compact layer, as previously noted in the chick ventricle (Sedmera et al., 1997). The individual trabeculae were relatively long and radial, with fewer branching points, and less organized than in the embryogenesis of higher vertebrates. The semicircular trabecular folds reinforced the trabeculated ventricle, especially the trabecular-free central lumen during the ventricular pumping. The bulboventricular valve was located at the anterior portion of the ventricle and had two semilunar valve cusps, with one on the left side of the ventricle, and the other on the right (Figs. 2, 7). The valve orifice (slit) was perpendicular to the dorsal plane of the ventricle, and the cusps were not attached, thus, did not form the commissures.
Atrioventricular orifice was positioned medially on the dorsal side of the anterior portion of the ventricle, almost directly above the bulbus arteriosus, and connected to the single-chambered atrium (Figs. 2A, 3). A fibrous cylindrical bridge joined the atrium to the ventricle. The size of the atrium could be larger than the ventricle when fully extended. Pectinate muscle radiated throughout the internal atrial wall from near the atrioventricular orifice to the opposite wall of the atrium (Fig. 2A). The compact layer at the atrioventricular orifice was the basal leaflets of the atrioventricular valve, which were free from tension apparatus. There were trabecular bands anchoring to the atrioventricular junction, and papillary muscle supporting the atrioventricular valve leaflets (Fig. 2C). The atrioventricular valve had four distinct leaflets oriented anterior, posterior, left, and right of the atrioventricular orifice (Fig. 3).
Although ventricular myocytes were larger than those in the atrium, the relative proportion of myofibrils was similar to the atrium, which occupied more than half of the myocyte volume (Table 1). Percent mitochondria in the myocytes were also similar in both atrium and ventricle. Myofibrils were relatively centrally located in the cell, with mitochondria close to the peripheral of the sarcolemmal (Fig. 4A,B). The trabeculae lacked coronary supply; thus, the myocytes contained abundant mitochondria presumably to maximize energy delivery to the contractile apparatus. Gap junctions with desmosome between cells were frequently found in both atrium and ventricle. Sarcoplasmic reticulum was sparse in zebrafish ventricle, and the sarcolemmal lacked a t-tubule system.
Table 1. Myocyte size and intracellular proportion of organelles in atrium and ventricle of the adult zebrafisha
Values are expressed as mean ± SEM.
15.19 ± 1.52 μm2
22.04 ± 1.39 μm2
59.2 ± 4.4%
55.6 ± 3.6%
24.4 ± 2.1%
28.7 ± 1.8%
Oxygenated blood reached the ventricle from the efferent branchial arch arteries through the coronary vessel on the ventral surface of the bulbus arteriosus (Fig. 5). The coronary vessels formed a visible epicardial network (Fig. 3) and branched into the compact layer (Fig. 2B). The ultrastructure of the coronary vessel conformed to the appearance of the muscular artery in higher vertebrate, with a single layer of endothelial cells lined at the arterial lumen (Fig. 6). The media was composed of spirally arranged pericytes, and the outside elastic lamina is indistinct. The coronary capillaries ended at the subtrabecular layer near the compact layer. The lumen of the coronary vessel could be extremely small so that it facilitated only the size of a single blood cell. Grossly, there were more coronary arteries visible on the dorsal than the ventral surface of the ventricle. The compact layer was well vascularized with capillaries connecting to veins that drained into the atrial chamber close to the atrioventricular region.
The luminal wall in the bulbus arteriosus was a heavily ridged muscular vasculature running lengthwise to the pear-shaped chamber, and continuing directly to the smooth-walled ventral aorta (Fig. 2A). The bulbus arteriosus had three distinct layers: the outer (externa), middle (media), and inner layer (intima) (Fig. 5). The externa was composed of scattered layers of interrupted elastic lamina, followed by 7 to 10 layers of helically arranged smooth muscle surrounded by a fine network of distensible collagenous and reticular fibrils in the media. The intima was composed of subendothelium with longitudinally oriented collagen, reticular fibrils, elastic fibers, and smooth muscle cells bound by a thin endothelial layer. Collagen fibrils were in close association with the vasculature smooth muscle cells (Fig. 7).
The atrium-developed (systolic) pressure (0.68 ± 0.04 mmHg) was consistently higher than the ventricular end-diastole (0.42 ± 0.10 mmHg), producing a pressure gradient across the atrium and the ventricle. Atrial pressure was lowest at the onset of atrial filling that coincided with the onset of ventricular ejection (Fig. 8). Ventricular pressure tracing showed that the inflow of blood occurred early in atrial systole, and the ventricular diastole constituted almost two-thirds of the cycle length (397.7 ± 26.4 msec). With closure of the atrioventricular valve, ventricular pressure rose rapidly to the level (2.51 ± 0.27 mmHg) at which it opened the bulboventricular valve, and pumped the blood to the bulbus arteriosus. With the declining ventricular pressure (peak systole to end-systole), the bulboventricular valve closed. Ventricular maximum dP/dt occurred at the point of ventricular ejection, and the ventricular minimum dP/dt matched with the ventricle end-systole (Fig. 8). There was a peak systolic pressure gradient from the ventral (2.16 ± 0.21 mmHg) to the dorsal (1.51 ± 0.38 mmHg) aorta.
Zebrafish heart, as in avian and mammalian hearts, develops from the cardiac progenitor cells, which migrate medially to form the bilateral plate mesoderm (DeHann, 1965; Viragh et al., 1989; DeRuiter et al., 1992; Stainier et al., 1993). The heart tube starts beating rhythmically after the fusion of the two primitive myocardial tubes on either side of the midline to enclose the endocardial cells (Manasek, 1970). Subsequently, the heart tube transforms from a smooth walled cardiac loop to a mature trabeculated heart with distinct chamber of sinus venosus, atrium, ventricle, and bulbus arteriosus (Hu et al., 2000).
The form of the ventricle and the relative development of the compact layer with its associated coronary circulation are useful to categorize the zebrafish heart (Farrell and Jones, 1992). The ventricular mass is directly proportional to the increasing body mass in the developing zebrafish (Barrionuevo and Burggren, 1999; Hu et al., 2000). The larger relative ventricular mass in the adult zebrafish is important in developing high blood pressures to accommodate a large cardiac stroke volume and active metabolism, and to compensate for the negative inotropic effect of low temperature (Clark et al., 1986; Nakazawa et al., 1986).
The ventricular size and shape are essential in determining cardiac function, which are likely dependent on the Frank-Starling mechanism and Laplace's law. The function of a chamber depends on its hydrodynamic properties which include the isometric pressure exerted by the chamber distension, compliance of the chamber, and wall resistance within the chamber (Regen, 1988). The pyramidal shape of the zebrafish ventricle is advantageous to pressure development, as the ventricular apex has a small radius of curvature (Farrell and Jones, 1992). The dynamic interactions between the internal functional force of the cells, and the position of the cell lineage, subsequently determine the shape of the heart (Terracio and Borg, 1988).
The cardiac muscle cell structure of the pectinate muscle is similar to that of the myocardium elsewhere in the atrium. During atrial contraction, the pectinate muscle shortens and pulls the atrium down onto the ventricle and forces the blood into the ventricular chamber. The pectinate muscle is randomly striated, and its orientation on the left side is apparently different from the right side of the atrium. This finding suggests at least one other segment of the heart, in addition to D-looping in the ventricle, has left-right asymmetry.
The trabeculation in the ventricle has a pattern similar to that of the myoarchitecture we found in the chick (Sedmera et al., 1997, 2000). The trabeculae account for the greater proportion of ventricular mass and maintain the conformity of the heart during the systolic and diastolic motion. The trabecular arrangement is essential for force generation in the heart and supports the role of ventricular contractility (Icardo and Fernandez-Teran, 1987; Sanchez-Quintana and Hurle, 1987; Sedmera et al., 2000). In zebrafish, the ventricular chamber has a relatively thin compact layer, with extensive and elongated trabeculae. The long and slender trabecular mesh creates numerous tiny intertrabecular spaces (lacunae) that serve as “small heart components” within the large ventricle (Johansen, 1965). Because the size of each intertrabecular space is tiny, the heart increases its contractile efficiency and the pressure is generated at a considerable mechanical advantage compared with the heart consisting of smooth-walled myocardium (Hu et al., 1991; Taber et al., 1993).
The myocytes in the zebrafish ventricle are relatively small. The size of the myocyte in homeotherms is proportional to stroke volume and inversely proportional to heart rate and to the maximum activity state of the animal (Paoupa and Lindstrom, 1983). Mature zebrafish have a relatively fast heart rate (Barrionueva and Burggren, 1999; Hu et al., 2000). The fast heart rate in small mammals is characterized by small myocytes. The abundance of mitochondria at the periphery of cells increases the diffusion efficiency of oxygen to the organelles. The frequent occurrence of gap junctions and the abundance of myofibrils in the ventricle enhance the performance of the heart. With the absence of a t-tubule system which is typical in mammalian cardiac and skeletal muscles, calcium diffusion from the extracellular space to troponin-C can be facilitated, which is critical for cardiac function.
The appearance of the epicardium can readily be identified as early as 48 hr postfertilization when the zebrafish heart emerges as four distinct structure of sinus venosus, atrium, ventricle, and bulbus arteriosus (Hu et al., 2000). In adult zebrafish, the epicardium in the ventricle is formed by a single layer of mesothelial cells supported by a basal lamina, and imbricated with collagen, fibroblasts, and vascular structures in the subepicardial space. The mesothelial cells are interdigitated to each other by desmosomes to facilitate the mechanism of changes in heart volume during the cardiac cycle (Lemanski et al., 1975; Santer, 1985).
Coronary circulation derives cranially from the hypobranchial artery, which is a convergence from efferent branchial arteries (Farrall and Jones, 1992). Thus, the oxygenated blood is delivered to the heart through the gills in a direct manner. Before the coronary artery reaches the ventricle, it courses across the dorsal surface of the bulbus arteriosus. The path of the coronary artery from the efferent branchial arch arteries to the proximity of the bulbus arterious remains undefined. With confocal microangiography (Weinstein et al., 1995; Isogai et al., 2001), the route of the coronary vessels may possibly be identified in zebrafish. Because the coronary vasculature is retained mostly around the subepicardial space, the trabeculae may also derive nutrition and oxygen from luminal blood. As a result, this may maximize the cross-sectional area for tension development but minimize the diffusion distance for oxygen transfer from the outside to the center (Harper et al., 1993). The functional advantage of the long and slim trabeculae is a shorter diffusion distance from the outside to the center of the cell. However, such shape probably results in higher electrical resistance and slower conduction velocity.
The bulbus arteriosus of different species each presents unique and specific characteristics that make it different from the bulbus among other species of the same class (Icardo et al., 1999a, b, 2000). Both conus and truncus arise in early heart development. This finding supports the notion that phylogenetically the primitive conus has been condensed into the ventricle (VanPraagh and VanPraagh, 1966; Goor et al., 1972; Lev, 1972, Anderson et al., 1974; Clark et al., 1984). Some investigators suggest that the teleost bulbus is of cardiac origin (Yamauchi, 1980; Farrell and Jones, 1992). The bulbus arteriosus of the zebrafish is composed of cardiac myosin origin during early development but transformed to the smooth muscle phenotype, which is characterized in the vasculature (Hu et al., 2000). The MF20 antimyosin antibody staining shows that the positive/negative transition during early development may be attributed to the removal of outflow tract cardiomyocytes through apoptosis (Watanabe et al., 1998). It may also be a replacement of the smooth muscle cells from the head mesenchyme. We describe the organization of the mature bulbus arteriosus with the arterial terminology, because the structure of the bulbus arteriosus in the mature zebrafish more closely resembles vasculature. By far, zebrafish should only be described as a two-chambered heart with a single atrium and ventricle that pumps the venous blood to the distensible bulbus arteriosus.
Blood pressure is a major variable in the regulation of the zebrafish cardiovascular system. Zebrafish has a unidirectional flow through the inlet atrioventricular and outlet bulboventricular valves. Blood moves down the ventricular diastolic pressure gradient, which identifies the atrioventricular valve as a site of flow resistance that influences the ventricular filling (Hu and Keller, 1995). The bulbus arteriosus in the zebrafish is noncontractile, and it stretches within the physiological parameter, and simultaneously with each ventricular systole, but returns to its original state during the ventricular diastole. Thus, it functions as an elastic reservoir or “windkessel,” which absorbs the rapid rise of the ventricular pressures during the ventricular contraction, thereby smoothing the effect of systolic pressure on the delicate gills vasculature (Johansen, 1965; Milnor, 1989; Farrell and Jones, 1992). The elastic rebound of the bulbus arteriosus provides a constant flow into the gills, which is essential for adequate gas exchange.
There is a 40% pressure drop from the dorsal to ventral aorta. The pressure gradient is created by means of the resistance in the fine branchial arteries. The rate of the ventral aortic pressure decline, that is, the downward slope of the ventral aortic pressure waveform, is dependent on heart rate and peripheral resistance. Resistance is characterized by the relationship of pressure to the reciprocal of flow, the geometric features of the vessel, dimensions, and viscosity of the blood (Clark and Hu, 1982, 1990; Hu and Clark, 1989; Hu et al., 1991). A gradual-sloping pressure tracing (relaxation rate) indicated a slow heart rate and a steady run-off due to high peripheral resistance in the branchial arteries (Zahka et al, 1989; Hu et al, 1992; Yoshigi et al, 1996; Keller et al., 1997).
Structural remodeling and changes in the function can be caused by many factors, including changes in tissue metabolism, animal growth and movement, or environmental perturbations. The results of this study provide a fundamental description of the morphologic structure and physiologic function of the adult zebrafish heart. It serves as a framework for further studies on the morphofunctional design of the zebrafish cardiovascular system. Detailed studies in different aspects of structure and function in the zebrafish cardiovascular system are essential to further understanding the mechanisms of the integrative relation of morphology and function of normal and abnormal heart formation.
We thank Roberto A. Monge and Thom A. Jensen, Jr., for their excellent technical support, and Kelly Fadden for the final language revision. This study was supported by an Innovative Research Grant from Primary Children's Medical Center Foundation, Salt Lake City, Utah. H. Joseph Yost, Ph.D. is an Established Investigator of the American Heart Association.